Data pre-processing
PhysioDataToolbox (Version 0.6.3), a MATLAB-based (MATLAB 2020b
Component Runtime (v9.9)) application, was used to pre-process data from
Biopac (Sjak-Shie, 2022). All data were formatted and imported into the
toolbox, and then the ECG signal analyzer and HRV analyzer in the
toolbox were used. The ECG signal analyzer treated the raw ECG data with
a 1 Hz high-pass filter, a 50 Hz low-pass filter, and 1x signal gain.
Then the ECG analyzer detected R-peak with the feature of 0.5 mV minimum
R-peak, 0.3 s minimum distance between R-peak, 0.3 s minimum interbeat
interval value, and 2 s maximum interbeat interval value. Baseline
epochs with 5 or 10 minutes were defined according to the procedures of
different experiments. After the R-peaks detection, the ECG data were
visually inspected and manually corrected to remove ectopic beats,
artifacts, and misidentified R-peaks singly. We extracted the root mean
square of continuous heartbeat interval difference (RMSSD) to assess
heart rate variability in time domains for the baseline epoch. Although
some frequency-domain metrics such as high-frequency power also reflect
parasympathetic activity, RMSSD is more correlated with vagal regulation
and is less affected by respiratory and motor artifacts (Penttilä et
al., 2001). The data from Polar, which only measures the intervals
between two R-peaks in a consecutive period, were different from the
data from Biopac which contains the whole heartbeat cycles and
intervals. As such, we analyzed these data via Artiifact and Kubios
3.0.2 (Kaufmann et al., 2011; Tarvainen et al., 2002, 2014). The
published polar data were adopted directly from three studies from our
lab (Pulopulos et al., 2020a; Pulopulos et al., 2020b).